Search site

Estimating changes in river flow in the UK

Objectives: CEH used UKCP09 data to undertake a study that sought to estimate changes in river flow across the river network of Britain.

UKCP09 products used: 11-member RCM

How were UKCP09 products used?

  1. Data from the 11-member Regional Climate Model (RCM) for baseline and future time periods and for all the ensemble members, were downscaled to a 1 km resolution (the required resolution) and input into a Grid-to-Grid hydrological model. The downscaling procedure used high-resolution information from a standard average annual rainfall dataset.
  1. A change in river flow from baseline to future was calculated on a 1 km grid across the UK for each ensemble member.
  1. A map showing estimated changes in peak annual average flow across the UK was developed for each ensemble member projection.

Difficulties & limitations

A limitation of the 11-member RCM outputs is the extent to which they (and the 11 sets of results produced) explore uncertainties in the modelling process. The spread of results provided by the 11 regional climate models does provide some quantification of the uncertainty associated with the Hadley Centre climate model. However, unlike the UKCP09 probabilistic climate projections, the 11-member RCM output does not incorporate the results of other global climate models or include any consideration of relative likelihood or probabilities of certain outcomes.

Furthermore, 11-member RCM output is only available for a single emissions scenario (A1B; UKCP09 medium) meaning that the effect of emissions uncertainty is not addressed.

The main difficulty encountered was the sheer volume of data. Sixty years worth of hourly data required 3.6 GB of storage for one variable for each ensemble member. Thus our data storage requirement was approximately 200 GB just for the input data for our models.

It is also important to note that the climate models assume a year contains 360 days (12 months of 30 days). The viewing program  xconv is a quick and easy way of looking at the data-files.

Lessons learned

The main learning curve related to the use of an ensemble of RCMs, giving us a set of 11 possible future changes in river flows. In previous work (Bell et al. 2007) there were just one set of results to analyse, which provided an indication of how river flows may change under a plausible future climate. Here, additional expertise will be required to judge the relative merits of the projected changes from the 11 regional climate models in order to provide some guidance on the likelihood of the resulting projected changes in river flows.

How will the results be communicated to the target audience?

The target audience consists of customers such as Defra and the Environment Agency for whom results are communicated through reports and publications in refereed scientific journals.

Combining all the river flow results from 11 RCMs has provided a challenge; at present there is little guidance about which of the 11 regional climate models are most reliable. The full range of regional climate model output has been illustrated (see Figure 2 in the full report), and then presented the mean of all the RCM-derived river flow results alongside the full range of results.

Find out more

  • Contact details: V.A. Bell, Centre for Ecology and Hydrology, Richard Jones, Met Office Hadley Centre (Reading).